Error correction stops being theory
Logical qubit counts moved from single digits to around a hundred. That does not make a useful machine, and it changes the timetable for everything downstream.
Daniel Miksha via UnsplashQuantum computing has always had two numbers. The one in press releases is physical qubits. The one that decides whether a machine can do anything is logical qubits, meaning qubits stable enough to hold an answer through a computation.
That second number has moved. Summaries of the year's milestones describe multiple vendors reporting logical qubit counts approaching a hundred, with error rates falling as more physical qubits are added rather than rising.
The direction of that relationship is the whole result. For years adding qubits added noise faster than it added capability. Exponential error suppression means the engineering now works with the physics rather than against it.
None of this produces a useful machine yet, and the vendors are unusually clear about that. Published roadmaps put the first large fault-tolerant systems toward the end of the decade, at scales an order of magnitude beyond today.
The gap between those two facts is where the commercial confusion lives. A genuine scientific milestone is being marketed as imminent capability, and buyers are being asked to plan for something that does not exist.
The one decision that does need making now is unrelated to whether anyone runs a quantum programme. Data encrypted today with current public key cryptography can be captured and stored, and decrypted later by a machine that does not exist yet.
That makes migration to post-quantum algorithms a present-tense project with a long tail, and it is driven by the sensitivity lifetime of the data rather than by the arrival date of the hardware.
For everything else, the honest posture is attention rather than investment. The applications most often cited, in chemistry and materials, are real and require machine scales that current roadmaps place years away.
Cloud access has made experimentation cheap enough that there is no reason not to look, and cheap enough that looking should not be confused with a strategy.
What has actually changed is confidence rather than capability. A field that could not previously say the engineering path was viable can now say so, which moves the risk from whether to when.